Event ·Multi-source

LlamaIndex 推出 Extract Max Effort,文档提取达 ExtractBench Pro 61% 准确率

LlamaIndex 为 LlamaParse 新增 Extract Max Effort 模式,针对复杂文档按页与 schema 复杂度动态分配 token 进行推理。该模式在 ExtractBench Pro 上取得 61% 的完美提取准确率,价格表现优于 Codex+Astra 与 Claude Code+Opus 5.5。适用场景包括数百页的保险费率文件、测井图表和建筑蓝图。该功能是在 Agentic Plus 模式上的一个开关,按额外计算量计费。

Current read

LlamaParse 出了个 Max Effort 模式,几百页保险文件、测井图都能抠出来,比 Opus 5.5 还便宜,做数据提取的可以试试。

11 sources62° AI heatLast updated 2026/11/1 17:40:15

Evidence

11 sources
Related source 1rohanpaul_ai
View source
Related source 2Simon Willison’s Weblog
View source
Related source 3kimmonismus
View source
Related source 6@OpenAIDevs
View source
Related source 7TestingCatalog
View source
Related source 10thsottiaux
View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.